Abstract
In this work, we propose and investigate a strategy to prevent consensus in kinetic models for opinion formation. We consider a large interacting agent system and assume that agent interactions are driven by compromise as well as self-thinking dynamics and also modulated by an underlying static social network. This network structure is included using so-called graphons, which modulate the interaction frequency in the corresponding kinetic formulation. We then derive the corresponding limiting Fokker–Planck equation and analyze its large time behavior. This microscopic setting serves as a starting point for the proposed control strategy, which steers agents away from mean opinion and is characterized by a suitable penalization depending on the properties of the graphon. We show that this minimalist approach is very effective by analyzing the quasi-stationary solution mean-field model in a plurality of graphon structures. Several numerical experiments are also provided to show the effectiveness of the approach in preventing the formation of consensus steering the system toward a declustered state.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
Opinion formation models have been extensively studied in several research communities in the last decades. Classical models are based on the assumption that an individual’s opinion is influenced by binary interactions with others as well as their surrounding (for example, through social media). Most of them describe the dynamics of each individual, resulting in complex large systems, see, for example, DeGroot (1974), Amblard and Deffuant (2004), Hegselmann and Krause (2002), Galam (2008), Iacomini and Vellucci (2023), Motsch and Tadmor (2014), Sznajd-Weron and Sznajd (2000), Goddard et al. (2021), Jabin and Motsch (2014). In many of these models, the underlying microscopic dynamics lead to the formation of complex macroscopic patterns and collective states. Methodologies from statistical mechanics, especially kinetic theory, have been used successfully to derive and analyze these complex stationary states in suitable scaling limits. Toscani’s seminal work on kinetic opinion formation models, see Toscani (2006), was one of the starting points of various kinetic models for collective dynamics, studying, for example, the effects of leaders on the opinion formation process (Albi et al. 2014; Düring et al. 2009; Düring and Wolfram 2015), decision making (Pareschi et al. 2017, 2019) and the influence of exogenous factors (Bondesan et al. 2024; Franceschi et al. 2023; Zanella 2023) to name a few.
There has been a general agreement that individuals change their opinion due to interactions with others. (These interactions are almost always assumed to be binary.) Most models assume that only like-minded individuals interact, known as bounded confidence models, and that dynamics are strongly driven by the tendency to compromise. In addition, they often assume that individuals change their opinion due to self-thinking, for example because of exposure to different media channels. There is a rich literature on models for opinion formation in large interacting agent systems, see, for example, Hegselmann and Krause (2002), Borra and Lorenzi (2013), Motsch and Tadmor (2014) as well as the influence of social networks on the opinion formation process. The later trend was accelerated as more and more data from social networks, such as X (formerly known as Twitter) or Facebook became available. This allowed researchers to investigate, for example, challenging questions related to the influence of voter’s behavior on the success of vaccination campaigns, see, for example, Albi et al. (2024, 2017). For other related works about the control of opinions on evolving networks, see Albi et al. (2016), polarization see Lee et al. (2014), Matakos et al. (2017), Amelkin et al. (2017) and marketing aspects see Toscani et al. (2018).
Several works proposing different control strategies to enhance consensus formation can be found in the multi-agent system literature, see, for example, Carrillo et al. (2010a, 2010b). On the other hand, control strategies to prevent consensus formation, we will refer these strategies as declustering, have been less studied (Piccoli et al. 2019). However, these declustering strategies can be a useful tool to understand how, for example, software-managed social media accounts, also known as bots, can prevent consensus formation or steer opinions in social networks; see, for example, Alothali et al. (2018), Gilani et al. (2019). It is therefore of interest to understand control mechanisms, which prevent consensus formation in large social networks.
Social networks can be studied using graph theory. Graph theory has become one of the most active fields of research in connection with the collective behavior of large populations of agents, see, for example, Barabási and Albert (1999), Barabási (2009), Barré et al. (2017, 2018), Newman (2003), Watts and Strogatz (1998). The necessity to handle millions, and often billions, of vertices lead to the study of large-scale statistical properties of graphs. In recent years, large discrete networks have been treated as continuous objects through the introduction of new mathematical structures called graphons, which stands for “graph functions”, see Borgs et al. (2014), Lovász (2012), Van Der Hofstad (2016), Caron et al. (2023), Glasscock (2015). The main feature of graphons relies on the possibility to bypass the introduction of classical adjacency matrix by looking at an associate function W(x, y) encoding all the information on the connectivity of the original discrete graph. Graph-theoretical research aside, the concept of graphons has been also applied to optimal control (Gao and Caines 2019; Hu et al. 2023) and especially epidemiological theory (Naldi and Patane 2022; de Dios et al. 2022; Erol et al. 2023). We mention the recent works on kinetic and mean-field equations on graphons (Bonnet et al. 2022; Bayraktar et al. 2023; Coppini 2022; Nurisso et al. 2024): In particular, graphons are leveraged to analyze the behavior of mathematical models acting on networks when their number of nodes becomes very large. Thorough analysis of such limiting procedures has been also carried out recently—see, e.g., Medvedev (2014), Petit et al. (2021) and references therein—increasing the interest on connections between differential equations models as continuous limits of stochastic processes happening on discrete graphs, like random walks and diffusion processes, for example.
Fokker–Planck-type equations acting on suitable limits of large dense graphs in particular have been receiving increasing attention (Coppini 2022; Bayraktar et al. 2023). Within this framework, our analysis is devoted to a continuous model acting on a very large network (not necessarily dense) via its graphon representation rather than its limiting sequence of finite graphs. This allows for the very natural interpretation, from the kinetic theoretical point of view, of the network as a (continuous) interaction kernel, which then leads to an effective surrogate mean-field model that incorporates the graphon.
In this paper, we propose and investigate possible strategies to prevent consensus formation in a kinetic model for opinion formation on networks. Our main contributions are
-
Development and analysis of a minimal control problem on the agent-based level.
-
Derivation of a closed form solution of the controlled model in suitable scaling limits, showing that the proposed strategy does indeed prevent consensus in the long time limit (under certain conditions on the parameters).
-
Provide extensive computational experiments illustrating the effectiveness of the proposed control strategy for power-law, \(k\)–\( NN \) and small-world graphons.
This paper is structured as follows: In Sect. 2, we introduce and analyze a kinetic opinion formation model on a stationary network. In Sect. 3, we propose a simple but very effective control mechanisms, which prevents the formation of consensus. In particular, we are able to provide closed form optimal controls, which prevent consensus formation in certain parameter regimes. We corroborate our analytical results with computational experiments in Sect. 4.
2 Kinetic Models for Opinion Dynamics on Graphon Structures
We consider a large population of indistinguishable agents each characterized by their opinion w belonging to \(I:=[-1, 1]\), where \(\pm 1\) corresponds to two opposite believes. Agents change their opinion through binary interactions with an interaction frequency modulated by an underlying static network. The opinion formation process itself is based on two different mechanisms:
-
first compromise dynamics—so individuals with close opinion try to find a compromise
-
and second opinion fluctuation, which are included via random variables.
The interaction frequency of agents depends on the underlying graph structure, which we model by a graphon in this paper.
Before discussing the binary agent interactions, we recall some basics about graphons (and refer the interested reader to “Appendix A” for a more detailed introduction as well as further references). Graphons are continuous objects that generalize the concept of simple graphs with a large number of vertices. In case of discrete graphs, nodes are usually referred to using the index \(i = 1, 2,\ldots , N\), where N is the number of vertices. However, in graphons the discrete set \(\{1,\ldots , N\}\) is mapped onto the continuous interval [0, 1], so that nodes are labeled as \(x \in \Omega \subseteq [0,1]\), where \(\Omega \) is a suitable subset of the unit interval of \(\mathbb R\). We will therefore consider agents which are not only characterized by their individual opinion w on a topic, but also their static position on the graphon \(x\in \Omega \subseteq [0,1]\).
We consider the following setup for binary interactions: Given two interacting agents characterized by their opinion and position in the graphon, that is, \((x,w),(y,w_*) \in \Omega \times I\), we compute their post-interaction opinions \((x,w^\prime ),(y,w_*^\prime )\) (note that their positions x and y in the graphon did not change) as:
where \(\gamma \in (0,1)\) is the so-called compromise parameter. The interaction function \(P(\,\cdot \,,\,\cdot \,) \in [0,1]\) depends on the graphon coordinates \(x,y \in \Omega \) and may also depend on the opinion variables \(w,w_* \in I\). For instance, in Toscani (2006) the case \(P = P(|w|)\) as a non-increasing function with respect to \(|w|\) and such that \(0 \le P(|w|) \le 1\) is explored. A different choice, resembling the case of a bounded confidence model like Hegselmann–Krause’s Hegselmann and Krause (2002), can be obtained setting \(P = P(|w - w_*|\)). Moreover, we remark that this particular choice has the advantage of ensuring the conservation of the average opinion but presents analytical difficulties arising from the presence of the absolute value. In the rest of the manuscript, we focus on the case where the interaction function only depends on the agents’ positions on the graphon. A possible choice could be
where \(d_i(z)\) is the in-degree of the node at coordinate \(z\in \Omega \) as defined in Definition 2 of Appendix A. Hence, interactions depend on the connectivity of each agent—the more connected an agent is the less it is influenced by the other, while agents with a lower connectivity are affected more. In particular,
- \(d_i(x)/d_i(y) \gg 1\):
-
implies that, on average with respect to the random variable \(\eta \), the agent with the highest degree keeps their opinion;
- \(d_i(x)/d_i(y) \approx 1\):
-
implies that agents with a similar number of incoming connections are the ones that can most influence each other;
- \(d_i(x)/d_i(y) \ll 1\):
-
implies that the less influential agents tend to adopt the opinion of the more connected ones.
A different possible choice of P(x, y) that would give rise to similar dynamics would be, for instance, \(P(x,y) = (1 + d_i(x)/d_i(y))^{-\alpha }\), with \(\alpha > 0\).
In Eq. (1), \(\eta \) and \(\tilde{\eta }\) are independent and identically distribution centered random variables with finite moments up to order three and such that \(\langle \eta ^2\rangle = \langle \tilde{\eta }^2\rangle = \sigma ^2 < +\infty \). Here we denote by \(\langle \,\cdot \,\rangle \) the expectation with respect to the distribution of the random variables. The variables \(\eta \), \(\tilde{\eta }\) account for random fluctuations in an individual’s opinion due to, for example, media exposure. The function \(D(\,\cdot \,, \,\cdot \,)\ge 0\) encodes the local relevance of the diffusion, possible choices include \(D(w) = \sqrt{1-w^2}\). In this case, agents diffuse the most if they have an indifferent opinion, that is, \(w \approx 0\), while they are less influenced by external factors once they settled on one of the two ‘extreme’ choices. Note that this choice also ensures that opinions stay within \(I\).
Next we discuss some basic properties of the binary interaction defined before. Under the assumption that the compromise propensity function P satisfies \( 0 \le P(x,y) \le 1\) and \(0 < \gamma \le 1/2\), the following Proposition (see, e.g., Pareschi et al. 2019; Toscani 2006; Toscani et al. 2018) ensures that the post-interaction opinions still belong to the reference interval.
Proposition 1
Assuming \(0 \le P(x,y) \le 1\) and
then the binary interaction (1) preserves the interval and the post-interaction opinions are such that \(w^\prime \), \(w_*^\prime \in I\).
A direct computation shows that for all w, \(w_* \in I\)
If \(P(x,y) = P(y,x)\), i.e., the compromise function P, is symmetric, mean opinion is preserved in interactions, that is,
On the other hand, the energy is not conserved on average since
If \(\sigma ^2 \equiv 0\) and we have symmetric interactions, we see that the mean energy is dissipated.
We can now state the evolution equation for the distribution of agents \(f = f(x, w, t) \) with respect to their position \(x \in \Omega \) and opinion \(w \in \mathcal {I}\). Consider a fixed number of players, N, then the binary interactions (1) induce a discrete-time Markov process with N-particle joint probability distribution \(P_N(x_1,w_1,x_2,w_2,\ldots ,x_N, w_N,t)\). This allows us to write a kinetic equation for the one-marginal distribution function,
using only the one- and two-particle distribution functions (Cercignani 2012; Cercignani et al. 1994),
Here, \(\langle \cdot \rangle \) denotes the mean with respect to the random variables \(\eta ,\tilde{\eta }\). By continuing this process, one obtains a hierarchy of equations, the so-called BBGKY hierarchy (Cercignani 2012; Cercignani et al. 1994), describing the dynamics of the system of a large number of interacting agents. It is a standard assumption to neglect correlations, implying that
By scaling time and performing the thermodynamical limit \(N\rightarrow \infty \), one can use standard methods of kinetic theory (Cercignani 2012; Cercignani et al. 1994) to show that the time evolution of the one-agent distribution function f is governed by the following non-Maxwellian Boltzmann equation
where Q(f, f) is the so-called collisional operator
Here \((x,{}' w)\) and \((y,{}' w_*)\) are pre-interaction opinions generating the post-interaction opinions (x, w) and \((y, w_*)\) and \({}' J\) is the Jacobian of the transformation \(({}' w,{}' w_*)\rightarrow (w,w_*)\). In Eq. (4), the kernel \({\mathcal {B}}(x,y) :\Omega ^2 \rightarrow \mathbb R^+\) is a given graphon. It can be thought of as the continuous equivalent of an adjacency matrix. Its use in (4) allows us to include an underlying network structure on the continuous level. Recent approaches to opinion formation modeling in the kinetic communities, e.g., Albi et al. (2024, 2016, 2017), Toscani et al. (2018), take into account a graph structure via some of its statistical descriptions, like, for example, considering the number of connections of each agents as an adjoint variable. Instead, the use of a graphon kernel allows for a richer and more general description of individual connections among agents.
Solutions to Eq. (3) preserve features of the underlying microscopic interaction rule. To compute the evolution of the mean opinion, we consider the weak formulation of Eq. (3). Let \(\varphi (x,w)\) be a test function, then
Setting \(\varphi (x,w)\equiv 1\) in (5) yields conservation of mass. Furthermore, from the conservation of the microscopic average opinion (2) we see that \(\varphi (x,w) = w\) gives conservation of the mean opinion (again assuming that the interaction function is symmetric).
Using conservation of the mean opinion, we can show that for any \(\varphi (x,w) = w^\alpha \phi (x)\), where \(\phi (\,\cdot \,)\) is a suitable test function, the macroscopic quantities
are conserved in time. Indeed, from (5) choosing \(\varphi (x,w) = \phi (x) w^\alpha \) we get
From the above equation, we get—for \(\alpha = 0,1\)—conservation of any weighted macroscopic moment of order \(\alpha \) defined in (6).
To investigate the second-order moment of f(x, w, t), we introduce the quantities
that is, the first- and second-order moment, respectively, with respect to w of agents with label \(x\in \Omega \). Clearly, integrating \(\Lambda (x,t)\) and \(\Xi (x,t)\) over \(\Omega \) gives us the mean opinion and the energy of the population.
We will see that for bounded graphons \({\mathcal {B}}(x,y)\) and no diffusion, that is, \(\sigma =0\), the agent distribution \(f(\,\cdot \,,w,t)\) converges toward a Dirac delta distribution centered in the initial mean opinion. This is not surprising since the opinion dynamics corresponds to a consensus formation process modulated by the graphon. Indeed, we have for all \(x \in \Omega \),
Therefore, in the uniform interaction case \(P(x,y)\equiv 1\), even in the presence of non-homogeneous graphon structure supposing \(\Vert {\mathcal {B}}(x,y) \Vert _{L^\infty (\Omega \times \Omega )}>0\), we get
Thus, we have that the second-order moment of \(f(\,\cdot \,,w,t)\) tends to its mean squared, and its variance vanishes exponentially fast. Integrating both sides with respect to \(x \in \Omega \) gives the estimate
where
are the second-order moment for the whole population and the global mean opinion, respectively. The latter quantity, thanks to Eq. (6), is conserved in time. Therefore, a bounded graphon kernel implies exponential convergence of the distribution f(x, w, t) toward a Dirac’s delta centered at the initial mean opinion.
2.1 Derivation of a Mean-Field Description
Since large time statistical properties of the introduced kinetic model are very difficult to obtain, several reduced complexity models have been proposed. In this direction, a deeper insight into the large time distribution of the introduced kinetic model can be obtained in the quasi-invariant regime presented in Toscani (2006). The idea is to rescale both the interaction and diffusion parameters making the binary scheme (1) quasi-invariant. The idea has its roots in the so-called grazing collision limit of the classical Boltzmann equation, see Düring et al. (2009), Pareschi and Toscani (2013) and the references therein. The resulting model has the form of an aggregation–diffusion Fokker–Planck-type equation which is capable of encapsulating the information of microscopic dynamics and for which the study of asymptotic properties is typically easier.
We consider \(\epsilon \ll 1\) and introduce the following scaling
for which \(w'\approx w\) and \(w_*' \approx w_*\). Next we Taylor expand the term encoding the binary interactions in the weak form of the collision operator of Eq. (5).
with \({\tilde{w}} \in (\min \{w,w'\},\max \{w,w'\})\). Introducing the new time variable \(\tau = \epsilon t\) and the corresponding rescaled density \(g(x,w,\tau ) = f(x,w,\tau /\epsilon )\), we can rewrite (5) as
where
and \(R_\varphi (f,f)/\epsilon \rightarrow 0\) under the hypothesis \(\langle |\eta |^3\rangle <+\infty \), see Cordier et al. (2005), Toscani (2006). Consequently, in the limit \(\epsilon \rightarrow 0^+\) we get
Therefore, integrating back by parts, we formally obtained a Fokker–Planck equation for the evolution of the distribution \(g(x,w,\tau )\)
where
The operator \({\mathcal {K}}\) corresponds to the network-modulated compromise process, the operator \({\mathcal {H}}\) corresponds to the network-weighted density \(g(x,w,\tau )\). Equation (9) is complemented wit no-flux boundary conditions for all \(x \in \Omega \):
This choice of boundary conditions ensures that system (9)–(11) shares the same conservation properties as its microscopic kinetic counterpart. Indeed, we see that the mean opinion is conserved since
where we dropped for clarity the dependence on x, w and t for the operators \({\mathcal {K}}[g]\) and \({\mathcal {H}}[g]\). Furthermore, any macroscopic quantity of the form
is conserved in time.
2.2 Large Time Agent Distribution
Due to the presence of a general compromise propensity \(P(\,\cdot \,,\,\cdot \,)\), a closed solution to Eq. (9) is difficult to obtain. Nevertheless, under suitable assumptions on the graphon structure and on the diffusion function, we can write down a closed formulation for the large time agent distribution (9).
In the following, we restrict our analysis to the simplified situation where the interactions are homogeneous, i.e., \(P(x,y)\equiv 1\), and the diffusion function is defined as
We recall that this choice of diffusion function ensures that w stays within the domain \(I\). Furthermore, we suppose separability of the graphon \({\mathcal {B}}(x,y)\), which corresponds to
From the modeling point of view, this choice is coherent with relevant examples of graphon structures, like the graphon associated with the case of scale-free networks as proposed in Borgs et al. (2014). Indeed, in this case we have
satisfying the introduced separability assumption. Network structures that are found commonly in life and social sciences are often modeled using scale-free networks (Van Der Hofstad 2016; Barabási 2009; Barabási and Albert 1999), i.e., simple graphs whose degree distribution possesses fat tails.
From (13), we define the weighted mass and momentum as
Note that both quantities \(\rho \) and \(\mu \) are conserved in time.
Assuming relation (15) holds, the steady state \(g^\infty (x,w)\) of the Fokker–Planck model (9) satisfies the following equation
Due to mass conservation and definitions (16), we can simplify it further and obtain
For our particular choice of diffusion function, that is, (14), we can compute the steady state of \(g^\infty (x,w)\) explicitly, see Toscani (2006). In particular, setting \(\lambda = \sigma ^2/\gamma \), we get
which, as a function of the opinion, is a Beta distribution, weighted by the in-degree \(d_i(x)\) at \(x \in \Omega \) times a graphon-dependent constant \(C_{{\mathcal {B}}}\) which depends on the way the splitting \({\mathcal {B}}(x,y) = {\mathcal {B}}_1(x){\mathcal {B}}_2(y)\) is obtained and such that the right-hand side of Eq. (17) has unitary mass.
Model parameters appearing in the steady state allow to get insights into the shape and other characteristics of the equilibrium opinion distribution: For instance, when \(\mu = 0\), \(g^\infty \) is an even function of w, so that the population has a neutral opinion on average. On the other hand, the balance between the actions of compromise and self-thinking dynamics expressed by the parameter \(\lambda \) tells us that if the action of the self-thinking is much stronger than the compromise one, i.e., \(\lambda \gg 1\), then the tendency of the population would be to polarize at the extremes, tending to a mixture of Dirac’s deltas at the boundary points of \(I\). We refer the interested reader to Toscani (2006) for an in-depth analysis of the roles of interactions parameters on the equilibrium distribution.
2.3 Analytical Properties
We continue by discussing some analytical properties for solutions to (9). First we show that (9) preserves the \(L^1\) regularity. To this end, we may rewrite Eq. (9) as
Note that we will again consider a specific form of diffusion, that is, \( D(w)=\sqrt{1-w^2}\). Furthermore, we introduce the following quantities, dependent on the compromise propensity function
Note that \(\mu _P(\,\cdot \,, \tau )\) is well defined since we are considering P to be positive almost anywhere and the graphon \({\mathcal {B}}\) to be nonnegative, as well. Next, we take, for a given parameter \(\xi \), a regularized non-decreasing approximation of the sign function \({{\,\textrm{sgn}\,}}_\xi (\,\cdot \,)\), and then introduce the antiderivative of \({{\,\textrm{sgn}\,}}_\xi [g (x,w,\tau )](w)\) for every \(w \in I\) as the function \(|g (x,w,\tau )|_\xi (w)\), where we stress the dependence on these functions on the variable w. Now, let us fix \(x \in \Omega \), multiply each side of Eq. (18) and integrate with respect to w. This gives
where we used the boundary conditions (11). Since
we can substitute this expression and obtain
Now, the first integrand on the right-hand side vanishes as \(\xi \rightarrow 0^+\) if we integrate by parts one more time (since by construction \(\lim _{\xi \rightarrow 0^+} {{\,\textrm{sgn}\,}}_\xi [g (x,w,\tau )](w)g (x,w,\tau )= |g (x,w,\tau )|(w)\) for almost every \(w \in I\)). This leaves us with the second integrand, which is nonnegative since we chose a non-decreasing approximation of the sign function. Finally, since graphons are nonnegative by definition at all points in their domain, we conclude that
for all \(x \in \Omega \). This implies that an initial datum in \(L^1(I)\) would ensure that \(g (x,w,\tau )\in L^1(I)\) for all \(\tau > 0\).
Remark 1
The weak contractivity of the \(L^1\) norm with respect to the opinion also allows us to prove uniqueness of solutions to (18). The proof is based on contradiction; assume there exist two solutions \(g (x,w,\tau )\) and \(s(x,w,\tau )\) and evaluate the regularized modulus of their difference for each point \(x \in \Omega \). If we fix \(x \in \Omega \), then due to linearity with respect to w we have that \(g (x,w,\tau )- s(x,w,\tau )\) is a solution to Eq. (18), too. Therefore,
which implies that, at \(x \in \Omega \), \(g = s\) for almost all \(w \in I\) and \(\tau > 0\) since by construction we have \(g(x,w,0) = s(x,w,0)\) for all \(x \in \Omega \) and \(w \in I\). The claim then follows since \(x\in \Omega \) was chosen arbitrarily.
Remark 2
The weak contractivity of the \(L^1\) norm with respect to w (i.e., the norm is not increased in time) gives us as a corollary that the model (18) is positivity-preserving. The claim follows noting that its solution \(g (x,w,\tau )\) has a vanishing negative part if the initial datum is nonnegative. Indeed, we can express the negative part of \(g (x,w,\tau )\) via the regularization we introduced earlier, that is,
This way, if we integrate with respect to w, we have
thanks to the first boundary condition in (11). Then it holds
for all \(x \in \Omega \), which implies that the non-negativity of the initial datum is preserved by the model (18).
We conclude this section by extending the previous regularity result—if the initial datum is in \(L^{p}(I)\), \(p>1\) at \(x\in \Omega \) then the solution \(g \in L^p(I)\) at x for all \(\tau > 0\). We will use both the positivity-preserving and the \(L^1\) regularity of its solution in the following.
We note that we only show \(L^p\)-regularity for all \(p \ge 2\) since the result will also hold also for \(p \in (1, 2)\) due to the boundedness of the interval \(I\). The idea is to rewrite Eq. (18) and impose the associated no-flux boundary conditions in order to estimate the time evolution of \(\Vert g\Vert _{L^p}\) using integration by parts. The hypothesis on p to be greater or equal than 2 is needed since we will need to take the derivative of \(g ^{p-1} (x,w,\tau )\) under the integral sign, but as stated above it is not restrictive.
We define the right-hand side of (18) as \({\mathcal {Q}}(g,g)(x,w,\tau )\), that is,
Suppose that the following no-flux boundary conditions hold for all \(x \in \Omega \) and for all \(\tau > 0\)
Now let us multiply each side of Eq. (9) by \(g ^{p-1} (x,w,\tau )\) and integrate with respect to w
Now the goal is to show that \({\mathcal {T}}_2\) is non-positive, so that it can be ignored in the estimate, and then focus on \({\mathcal {T}}_1\). Starting with \({\mathcal {T}}_2\), we integrate by parts and use the second boundary condition in (19) to obtain
since \(w \in I\), \(g (x,w,\tau )\) is nonnegative and \(\rho _P(x,\tau )\) is nonnegative at all \(x \in \Omega \). Next we consider \({\mathcal {T}}_1\) and derive two different estimates for it. If we expand the derivative with respect to w under the integral sign, we have
On the other hand, we could as well integrate by parts and using the first boundary condition in (19) and get
If we now use the identity
to replace \({\mathcal {T}}_1\) as the appropriate convex combination of the two equations, we obtain:
Putting everything together, we deduce that
for a given \(x \in \Omega \). Then Gronwall’s lemma implies that if the initial datum belongs to \(L^p([-1, 1])\) at \(x \in \Omega \), then \(g (x,w,\tau )\in L^p([-1, 1])\) at x for all \(\tau > 0\).
3 Declustering: Preventing Consensus via Control Strategies
In this section, we focus on control strategies to prevent consensus. In particular, the one driven by the compromise process. This objective is different to more common optimal control strategies, which would, for example, steer the average opinion to a given target (Albi et al. 2017, 2015, 2016). To this end, we propose an additional interaction to prevent the formation of opinion clusters by enforcing a controlled interaction. In particular, we consider a convex combination of two updates weighted by the parameter \(\theta \in (0, 1)\) such that a fraction \(1-\theta \) of the population follows an opinion transition of the type (1), whereas a fraction of size \(\theta \) follows an opinion update given by a controlled interaction of the form
where \(u^*\) is an agent-based control arising from the solution of a suitable optimization problem and \(S(w) \ge 0\) a suitable selection function dependent on the opinion. In particular, the optimization problem focuses on the minimization of a suitable convex cost functional \({\mathcal {J}}\) on the set \({\mathcal {U}}\) of admissible controls, which in our case are those such that the post-interaction opinion \(w''\) stays within the interval \(I\) and has the form
The quantity \(w''_*\) appearing on the right-hand side of (21) is a virtual update which the functional \({\mathcal {J}}\) would be subject to: In fact, the optimal control \(u^*\) is the solution of the following optimization problem
where m denotes the average opinion of the population on the network at time t and \(\nu > 0\) is a regularization parameter. Notice that the actual update \(w''\) and the virtual update \(w''_*\) have an opposite effect on w. Solving the associated Lagrange multiplier problem
gives
and so, the resulting interaction is
In particular, in the rest of the paper we focus on the selection function
Remark 3
Thanks to the presence of the indicator function in (24), we can verify that \(w'' \in [-1, 1]\) and therefore that the controlled interaction is admissible.
We recall that we balance the two types of interactions: At a rate \(1-\theta \), agents update their opinion according to (1), at the rate \(\theta \) they interact with the external control (23). This yields a kinetic model whose right-hand side is convex combination of two non-Maxwellian operators
The role of the parameter \(\theta \in [0,1]\) is to model the frequency at which the different kinds of interaction take place: It can be thought as the percentage of automated users (e.g., bots programmed by a third party) on the network. The operator Q(f, f) is the same introduced in Eq. (4); the operator \(Q_u(f)(x,w,t)\), instead, encodes the controlled update of agents’ opinions as prescribed by the elementary interaction (23) and is therefore given by
for any test function \(\varphi (\,\cdot \,,\,\cdot \,)\).
3.1 Mean-Field Limit of the Controlled Model
In this section, we explicitly show how the introduced control is capable of breaking consensus on the mean-field level for suitable choice of the penalization. We proceed like we did in Sect. 2.1 to derive a more approachable mean-field limit of Eq. (25), using the same scaling
for a certain \(\kappa \in \mathbb R^+\). In this case, particular care is needed in treating the weak form (26) due to the presence of the indicator function in the interaction (23). Using Taylor expansion like we did in Sect. 2.1 yields
Here, \({\mathcal {K}}[g]\), \({\mathcal {H}}[g]\) and \(R_\phi (g,g)\) are the same operators as in Eqs. (10) and (8) in Sect. 2.1, while we denote
which can be shown to go to zero with computations analogous to the ones for \(R_\phi (g,g)\). We continue with the term \(\theta {\mathcal {A}}[g](x,w,\tau )/\epsilon \):
since in the limit \(\epsilon \rightarrow 0^+\) we have
where we recall that \(\kappa > 0\) and \(|m| \le 1\). Therefore, we obtain the following Fokker–Planck equation
where this time we define
Here \(d_i(x)\) is the in-degree of x as defined in Sect. A and \({\mathcal {H}}_\theta [g] :=(1 - \theta ){\mathcal {H}}[g]\). The associated boundary conditions which are necessary to perform the integration by parts are
Remark 4
The mean opinion of the population is preserved. Indeed, multiplying each side of Eq. (28) and then integrating by parts we get
thanks to Eqs. (29) and (12). The evolution of the second-order moment is trickier, due to the presence of general graphon kernel and the compromise propensity function. In case of the specific diffusion function \(D(\,\cdot \,,\,\cdot \,)\) (14), we can simplify the expression and obtain
where we note \(\Phi (x,\tau ) :=\int _Ig(x,w,\tau )\, dw\). Equation (31) is still quite general: Further insights into the trend of the second-order moment can be found in the specialized setting of Remark 6.
3.2 Effects of the Penalization Coefficient on the Quasi-Equilibrium Distribution
In the following, we will compute the quasi-equilibrium distribution of (28). This corresponds to solving
which we can write in closed form (due to the no-flux bc (30)). Then the quasi-equilibrium distribution \(f^\textrm{qe}\) is given by
where C is a normalizing constant and under the formal assumption that \({\mathcal {H}}[g](x,\tau ) D^2(x,v)\) can be taken to be nonzero almost everywhere at \(x\in \Omega \) and on \([-1, 1]\).
To compare the controlled case to the one obtained in Eq. (17), we focus on the case \(D(x,w) = \sqrt{1 - w^2}\), and having chosen a unitary compromise propensity function, we can write
where we define
If we fix\(x \in \Omega \), the quasi-equilibrium state of Eq. (32) is a Beta distribution with respect to the variablew. In fact, taking \(\theta = 0\) and a separable graphon kernel \({\mathcal {B}}(x,y)\) in Eq. (32) gives us precisely the steady state we found for the uncontrolled problem, given by (17).
We can exploit our knowledge of the quasi-equilibrium to influence the level of declustering of the system: Indeed, the opinion distribution is in its least clustered form when it is uniform, i.e., when \(\alpha _- = \alpha _+ = 0\). If we impose these constraints, we can solve them for the penalty term \(\kappa \) as a function of the network position\(x\in \Omega \) and of time\(\tau \), i.e., \(\kappa =\kappa (x,\tau )\). We obtain
which, under the further hypothesis that \({\mathcal {B}}(x,y) = {\mathcal {B}}_1(x){\mathcal {B}}_2(y)\) and \(P(x,y) \equiv 1\), simplifies to
Remark 5
The constraint \(m = 0\) in (33) is necessary to have a uniform distribution over a symmetric domain, while imposing \(\mu _1(x,\tau ) = 0\) is needed to have \(\alpha _- =0\) and \(\alpha _+ = 0\) simultaneously.
When \(g \approx U(\Omega \times [-1, 1])\), that is, the distribution is almost uniform over the domain \( \Omega \times [-1, 1]\), we have \(\rho _1(x,\tau ) \approx d_i(x)\), so that
Finally, we stress that the choices (33) and (34) that would appear in the controlled update (20) come from the analysis of a quasi-equilibrium state for the distribution\(g(x,w,\tau )\) which has been computed from the Fokker–Planck Eq. (28). This implies that the penalty term would be effective in achieving the declustering effect only for a parameter regime in which \(\epsilon \) is sufficiently small.
Remark 6
If we consider again the evolution of the second-order moment as in (31), assume that \(P(\,\cdot \,,\,\cdot \,) \in L^\infty (\Omega \times \Omega )\), and use the \(\kappa = \kappa (x,\tau )\) defined in Eq. (33), we obtain
This estimate can be further simplified if \(m = \Lambda (x,\tau ) = 0\) for all \(x\in \Omega \) (a condition needed to observe a centered uniform distribution):
In particular, by Gronwall’s inequality we have that whenever the graphon kernel is bounded the variance converges exponentially in time toward 1/3, since integrating both sides with respect to \(x\in \Omega \) gives
where 1/3 is the variance of the uniform distribution over \(\Omega \times I\).
4 Numerical Tests
We conclude by illustrating the declustering strategy with various computational experiments. We show first the consistency of the quasi-invariant limit of the controlled model (25) in the network-homogeneous case. In this case, we choose a uniform, constant graphon kernel \({\mathcal {B}}(x,y)\equiv c \in (0,1]\); in particular, \({\mathcal {B}}(x,y) \equiv 1\). In the second example, we check the quasi-invariant limit using the power-law graphon, which is separable, to model an interaction happening on a scale-free network. In the third experiment, we illustrate the dynamics for non-separable, graphon kernels. All tests were performed using direct simulation Monte Carlo methods for the Boltzmann Eq. (25); we refer to Pareschi and Toscani (2013), Pareschi et al. (2019) and references therein for further details.
We start describing our method by rewriting Eq. (25) in strong form as a sum of gain and loss parts:
We indicate with \(Q^\Sigma \) and \(Q_u^\Sigma \) the operators obtained replacing the graphon kernel \({\mathcal {B}}(x,y)\) with the approximated version \({\mathcal {B}}^\Sigma (x,y\), given by
where \(\Sigma \) is an upper bound for \({\mathcal {B}}(x,y)\) over \(\Omega ^2\). Whenever we consider an unbounded interaction kernel (e.g., the power-law case), we consider a suitable truncation for \(\Sigma \). If we now highlight the gain and loss parts of \(Q^\Sigma \) and \(Q_u^\Sigma \), we have
where we define
Then, we discretize the time interval [0, T] with time step \(\Delta t > 0\) and denote as \(f^n(x,w)\) the time approximation \(f(x,w,n\Delta t)\) to consider the forward Euler-type scheme
where we define
We remark that under the condition \(\Sigma \Delta t \le 1\), \(f^{n+1}\) is well defined as a probability density.
Table 1 presents all the parameters we used in our computational experiments. Moreover, we always fix \(D(x,w) = \sqrt{1 - w^2}\) as diffusion function and \(P(x,y) \equiv 1\) as compromise tendency function. Finally, we consider as initial distribution a state close to full consensus represented by a truncated Gaussian distribution over the interval \({\mathcal {I}}\) for all \(x \in \Omega \)
where \(u_0 = 0\), \(\sigma _0^2 = \frac{1}{10}\) and \(C>0\) is a normalization constant.
4.1 Consistency of the Mean-Field Model: The Network-Homogeneous Case
We take model (28) with \({\mathcal {B}}(x,y) \equiv 1\), which corresponds to a fully connected network, in which every node is adjacent to every other. Using the quasi-invariant scaling, we again approximate the dynamics of (25) by the one-dimensional Fokker–Planck equation
Note that we choose \(\kappa = \bar{\kappa }\), as in Eq. (34), which corresponds to the optimal scaling to ensure declustering. Figure 1 shows that as \(\epsilon \) approaches zero, the controlled update gets fully effective and the state relaxes toward a uniform distribution.
This is also testified by Fig. 2, where we report the profile of the distribution \(g(w,\tau )\) at time \(\tau = T\), with \(T=8\). We also report the evolution from \(\tau = 0\) to \(\tau = T\) of the entropy, computed as
We can see that when \(\epsilon \ll 1\) the entropy approaches the value \(\log (2)\), which corresponds to the entropy of the uniform distribution over the interval \(I\).
4.2 Consistency of the Mean-Field Model: The Power-Law Network Case
Next, we take model (28) with \({\mathcal {B}}(x,y) = 9/16(xy)^{-1/4}\), i.e., the power-law graphon. We recall that this special choice yields Eq. (9). Since the power-law graphon is separable, the optimal value for the penalty term is \(\bar{\kappa }\) of Eq. (34) as for the network-homogeneous case. Figure 3 shows the evolution of \(g(x,w,\tau )\) for different values of\(\epsilon \). Since the distribution depends on both the opinion and the network position, we illustrate \(g(x,w,\tau )\) at three instances in time in the first row, that is, \(\tau = 0\), \(\tau = T/2\) and \(\tau = T\), where this time we fix \(T = 32\). The second row of plots in Fig. 3 shows the opinion marginal \(\int _\Omega g(w,t)\, dx \) for different values of \(\epsilon \).
Figure 4 again shows for ease of viewing the power-law graphon kernel that we use in our simulations and the time evolution of the entropy, computed as
Again, we see that in the limit \(\epsilon \rightarrow 0^+\) the state reaches an uniform distribution over \(\Omega \times I= [0, 1]\times [-1, 1]\), since the power-law graphon is defined on the entire unit square\([0, 1]^2\).
4.3 Declustering on Non-separable Networks
The last computational experiments illustrate the dynamics in case of non-separable graphons: the \(k\)–\( NN \) graphon and the small-world graphon. The first one models the \(k\)–\( NN \) networks as described, e.g., in de Dios et al. (2022), Watts and Strogatz (1998), Van Der Hofstad (2016) and it is defined de Dios et al. (2022) as
where \(\chi (\,\cdot \,)\) is the indicator function and \(r, p \in (0, 1)\) are constant real numbers, while the small-world graphon (see, e.g., de Dios et al. 2022; Watts and Strogatz 1998; Van Der Hofstad 2016) is defined as
Figure 5 shows the surface plots for both graphon kernels for fixed values of \(r = 1/8\) and \(p = 3/4\). We consider model (28) with scaling parameter\(\epsilon =10^{-3}\) and let evolve in time until\(T = 8\), where for this test we considered the network- and time-dependent optimal penalty coefficient\(\bar{\kappa }_{x,\tau }\) as written in Eq. (33). As we can see, \(g (x,w,\tau )\) approaches a uniform distribution over both network topologies.
5 Conclusion
In this paper, we proposed a simple yet very efficient optimal control strategy to break consensus in a kinetic model for opinion formation on graphons. The proposed approach allows us to include complex microscopic features, such as social networks, on the continuum limit and understand the impact of simple declustering mechanisms.
In doing so, we investigate the uncontrolled and controlled models in the mean-field limit. We then investigate the large time behavior and are able to write down closed form solutions of the (quasi)-stationary agent distribution of the uncontrolled and controlled problem for certain choices of parameters. This formulation allows us to identify the necessary controls to prevent consensus and steer the crowd toward a uniform distribution. We corroborate our analytical results with computational experiments for various types of graphons. The numerical results confirm our theoretical findings and the success of the proposed declustering strategy. Extensions of the designed approach to include dynamic networks for fully nonlinear equations are actually under study and will be presented in future researches.
References
Albi, G., Choi, Y.-P., Fornasier, M., Kalise, D.: Mean field control hierarchy. Appl. Math. Optim. 76, 93–135 (2017)
Albi, G., Calzola, E., Dimarco, G.: A data-driven kinetic model for opinion dynamics with social network contacts. Eur. J. Appl. Math. 1–27 (2024)
Albi, G., Herty, M., Pareschi, L.: Kinetic description of optimal control problems and applications to opinion consensus. Commun. Math. Sci. 13(6), 1407–1429 (2015)
Albi, G., Pareschi, L., Zanella, M.: Boltzmann-type control of opinion consensus through leaders. Philos. Trans. R. Soc. A 372(2028), 20140138 (2014)
Albi,G., Pareschi,L., Zanella,M.: On the optimal control of opinion dynamics on evolving networks. In: Bociu, L., Désidéri, J., Habbal, A. (eds) System Modeling and Optimization. CSMO 2015. IFIP Advances in Information and Communication Technology, vol. 494 (2016)
Albi, G., Pareschi, L., Zanella, M.: Opinion dynamics over complex networks: kinetic modelling and numerical methods. Kinet. Relat. Mod. 10(1), 1–32 (2017)
Alothali, E., Zaki, N., Mohamed, E.A., Alashwal, H.: Detecting social bots on twitter: a literature review. In: 2018 International Conference on Innovations in Information Technology (IIT), pp. 175–180. IEEE (2018)
Amblard, F., Deffuant, G.: The role of network topology on extremism propagation with the relative agreement opinion dynamics. Phys. A Stat. Mech. Appl. 343, 725–738 (2004)
Amelkin, V., Bullo, F., Singh, A.K.: Polar opinion dynamics in social networks. IEEE Trans. Autom. Control 62(11), 5650–5665 (2017)
Barabási, A.-L.: Scale-free networks: a decade and beyond. Science 325(5939), 412–413 (2009)
Barabási, A.-L., Albert, R.: Emergence of scaling in random networks. Science 286(5439), 509–512 (1999)
Barré, J., Carrillo, J.A., Degond, P., Peurichard, D., Zatorska, E.: Particle interactions mediated by dynamical networks: assessment of macroscopic descriptions. J. Nonlinear Sci. 28, 235–268 (2018)
Barré, J., Degond, P., Zatorska, E.: Kinetic theory of particle interactions mediated by dynamical networks. Multiscale Model. Simul. 15(3), 1294–1323 (2017)
Bayraktar, E., Chakraborty, S., Wu, R.: Graphon mean field systems. Ann. Appl. Probab. 33(5), 3587–3619 (2023)
Bondesan, A., Toscani, G., Zanella, M.: Kinetic compartmental models driven by opinion dynamics: vaccine hesitancy and social influence. Math. Mod. Methods Appl. Sci. (2024)
Bonnet, B., Duteil, N.P., Sigalotti, M.: Consensus formation in first-order graphon models with time-varying topologies. Math. Mod. Methods Appl. Sci. 32(11), 2121–2188 (2022)
Borgs, C., Chayes, J.T., Cohn, H., Zhao, Y.: An \(L^p\) theory of sparse graph convergence I: limits, sparse random graph models, and power law distributions. Trans. Am. Math. Soc. (2014)
Borra, D., Lorenzi, T.: Asymptotic analysis of continuous opinion dynamics models under bounded confidence. Commun. Pure Appl. Anal. 12(3), 1487–1499 (2013)
Caron, F., Panero, F., Rousseau, J.: On sparsity, power-law, and clustering properties of graphex processes. Adv. Appl. Probab. 55(4), 1211–1253 (2023)
Carrillo, J.A., Fornasier, M., Rosado, J., Toscani, G.: Asymptotic flocking dynamics of the kinetic Cucker–Smale model. SIAM J. Math. Anal. 42(1), 218–236 (2010)
Carrillo, J.A., Fornasier, M., Toscani, G., Vecil, F.: Particle, kinetic, and hydrodynamic models of swarming. In: Naldi, G., Pareschi, L., Toscani, G. (eds.) Mathematical Modeling of Collective Behavior in Socio-Economic and Life Sciences. Modeling and Simulation in Science, Engineering and Technology, Birkhäuser, Boston (2010)
Cercignani, C.: The Boltzmann equation and its applications. In: Applied Mathematical Sciences. Springer, New York (2012)
Cercignani, C., Illner, R., Pulvirenti, M.: The mathematical theory of dilute gases. In: Applied Mathematical Sciences. Springer, New York, (1994). ISBN: 9780387942940. https://books.google.com/books?id=F5NINMV-9IsC
Coppini, F.: A note on Fokker–Planck equations and graphons. J. Stat. Phys. 187(2), 15 (2022)
Cordier, S., Pareschi, L., Toscani, G.: On a kinetic model for a simple market economy. J. Stat. Phys. 120(112), 253–277 (2005)
DeGroot, M.H.: Reaching a consensus. J. Am. Stat. Assoc. 69(345), 118–121 (1974)
de Dios, B.A., Dovetta, S., Spinolo, L.V.: On the continuum limit of epidemiological models on graphs: convergence results, approximation and numerical simulations. arXiv:2211.01932 (2022)
Düring, B., Markowich, P., Pietschmann, J.-F., Wolfram, M.-T.: Boltzmann and Fokker–Planck equations modelling opinion formation in the presence of strong leaders. Proc. R. Soc. A 465(2112), 3687–3708 (2009)
Düring, B., Wolfram, M.-T.: Opinion dynamics: inhomogeneous Boltzmann-type equations modelling opinion leadership and political segregation. Proc. R. Soc. A 471(2182), 20150345 (2015)
Erol, S., Parise, F., Teytelboym, A.: Contagion in graphons. J. Econ. Theory 211, 105673 (2023)
Franceschi, J., Pareschi, L., Bellodi, E., Gavanelli, M., Bresadola, M.: Modeling opinion polarization on social media: application to COVID-19 vaccination hesitancy in Italy. PLoS ONE 18(10), 1–26 (2023)
Galam, S.: Sociophysics: a review of Galam models. Int. J. Mod. Phys. C 19(03), 409–440 (2008)
Gao, S., Caines, P.E.: Graphon control of large-scale networks of linear systems. IEEE Trans. Autom. Control 65(10), 4090–4105 (2019)
Gilani, Z., Farahbakhsh, R., Tyson, G., Crowcroft, J.: A large-scale behavioural analysis of bots and humans on twitter. ACM Trans. Web (TWEB) 13(1), 1–23 (2019)
Glasscock, D.: What is. a graphon. Notices AMS 62(1), 46–48 (2015)
Goddard, B.D., Gooding, B., Short, H., Pavliotis, G.A.: Noisy bounded confidence models for opinion dynamics: the effect of boundary conditions on phase transitions. IMA J. Appl. Math. 87(1), 80–110 (2021)
Hegselmann, R., Krause, U.: Opinion dynamics and bounded confidence models, analysis, and simulation. J. Artif. Soc. Soc. Simul. 5(3), (2002)
Hu, Y., Wei, X., Yan, J., Zhang, H.: Graphon mean-field control for cooperative multi-agent reinforcement learning. J. Frank. Inst. 360(18), 14783–14805 (2023)
Iacomini, E., Vellucci, P.: Contrarian effect in opinion forming: insights from Greta Thunberg phenomenon. J. Math. Sociol. 47(2), 123–169 (2023)
Jabin, P.-E., Motsch, S.: Clustering and asymptotic behavior in opinion formation. J. Differ. Equ. 257(11), 4165–4187 (2014). (ISSN: 0022-0396)
Lee, J.K., Choi, J., Kim, C., Kim, Y.: Social media, network heterogeneity, and opinion polarization. J. Commun. 64(4), 702–722 (2014)
Lovász, L.: Large Networks and Graph Limits, vol. 60. American Mathematical Society (2012)
Matakos, A., Terzi, E., Tsaparas, P.: Measuring and moderating opinion polarization in social networks. Data Min. Knowl. Disc. 31, 1480–1505 (2017)
Medvedev, G.S.: The nonlinear heat equation on dense graphs and graph limits. SIAM J. Math. Anal. 46(4), 2743–2766 (2014)
Motsch, S., Tadmor, E.: Heterophilious dynamics enhances consensus. SIAM Rev. 56(4), 577–621 (2014)
Naldi, G., Patane, G.: A graph-based modelling of epidemics: properties, simulation, and continuum limit. arXiv:2208.07559 (2022)
Newman, M.E.J.: The structure and function of complex networks. SIAM Rev. 45(2), 167–256 (2003)
Nurisso, M., Raviola, M., Tosin, A.: Network-based kinetic models: emergence of a statistical description of the graph topology. Eur. J. Appl. Math. 1–22 (2024)
Pareschi, L., Toscani, G.: Interacting Multiagent Systems: Kinetic Equations and Monte Carlo Methods. OUP, Oxford (2013)
Pareschi, L., Toscani, G., Tosin, A., Zanella, M.: Hydrodynamic models of preference formation in multi-agent societies. J. Nonlinear Sci. 29, 2761–2796 (2019)
Pareschi, L., Vellucci, P., Zanella, M.: Kinetic models of collective decision-making in the presence of equality bias. Phys. A Stat. Mech. Appl. 467, 201–217 (2017)
Petit, J., Lambiotte, R., Carletti, T.: Random walks on dense graphs and graphons. SIAM J. Appl. Math. 81(6), 2323–2345 (2021)
Piccoli, B., Duteil, N.P., Trélat, E.: Sparse control of Hegselmann–Krause models: black hole and declustering. SIAM J. Control Optim. 57(4), 2628–2659 (2019)
Sznajd-Weron, K., Sznajd, J.: Opinion evolution in closed community. Int. J. Mod. Phys. C 11(06), 1157–1165 (2000)
Toscani, G.: Kinetic models of opinion formation. Commun. Math. Sci. 4(3), 481–496 (2006)
Toscani, G., Tosin, A., Zanella, M.: Opinion modeling on social media and marketing aspects. Phys. Rev. E 98(2), 022315 (2018)
Van Der Hofstad, R.: Random Graphs and Complex Networks, vol. 43. Cambridge University Press, Cambridge (2016)
Watts, D.J., Strogatz, S.H.: Collective dynamics of small-world networks. Nature 393(6684), 440–442 (1998)
Zanella, M.: Kinetic models for opinion dynamics in the presence of opinion polarization. Bull. Math. Biol. 85(36), (2023)
Acknowledgements
This work have been written within the activities of the Royal Society Grant for International Exchanges Ref. IES/R3/213113. M.Z. and J.F. acknowledge the support of MUR-PRIN2020 Project No.2020JLWP23 and of the GNFM group of INdAM (National Institute of High Mathematics). M.Z. acknowledges the support of Next Generation EU. M.-T.W. acknowledges partial support from the EPSRC Small Grant EPSRC EP/X010503/1.
Funding
Open access funding provided by Università degli Studi di Pavia within the CRUI-CARE Agreement.
Author information
Authors and Affiliations
Corresponding author
Additional information
Communicated by Christian Bick.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Basic Definitions on Graphs and Graphons
Basic Definitions on Graphs and Graphons
We follow here Borgs et al. (2014), Glasscock (2015) to give a brief overview on graphs and graphons.
A (simple, unweighted) graph G is a pair of sets: V(G) that indicates vertices, or nodes, of G and E(G) that refers to the edges between nodes, pair-wise distinct. Two connected vertices are also said to be adjacent, and in this spirit, we can describe the graph G via its associated adjacency matrix \(A_{ij}(G)\), where \(A_{ij} = 1\) if and only if \((i,j) \in E(g)\), that is, if nodes i and j are connected. A way to represent the adjacency matrix of a graph is its pixel picture, i.e., we discretize the unit square \([0,1]^2 \subset \mathbb R\) into a grid of \(N^2\) squares of size 1/N, where N is the number of nodes of G. Then the square whose northwest coordinate is \(\bigl ((i-1)/N, (j-1)/N\bigr )\) is painted black if and only if \(A_{ij}(G) = 1\). Since matrices are labeled starting from their upper left corner, the same happens for the pixel pictures, where the origin is at their upper left corner as well.
If we let the number N become large, we see that the pixel picture “converges” toward a grayscale image. For example, if we consider a random graph where each node is linked to another with probability 1/2 seems to approach the constant function 1/2 over \([0,1]^2\).
This concept of limit of a sequence of graphs can be made rigorous introducing the notion of graphon.
We define a (labeled) graphon as a Lebesgue-measurable function from \(\Omega ^2\subseteq [0,1]^2\) to \(\mathbb R^+\), with the tacit assumption that we identify graphons that are equal almost everywhere. An unlabeled graphon is a labeled graphon to which is applied an invertible, measure preserving map to [0, 1] (called re-labeling).
A familiar example of a graphon is just a finite graph, so that it is sensible to see graphons as a consistent generalization of graphs. Indeed, finite graphs can be described as step functions on \([0,1]^2]\), i.e., measurable functions that are piecewise constant. We can construct the step graphon\(W^G\) associated with the finite graphG directly, pretty much like we did to pass from adjacency matrix to pixel picture:
where \(I_i \subseteq [0,1]\) is an interval of length \(1/|V(G)|\).
Graphons themselves become elements of a metric space (which we refer to as ) once we consider a distance between pairs (W, U) of them, called the cut distance
where \(\varphi \) and \(\psi \) are re-labeling while S and T are measurable subsets of \(\Omega \subseteq [0,1]\). Notice that the cut distance is just a pseudo-metric, since .
There are two main results that are useful when dealing with graphons Glasscock (2015), which we report for convenience.
Theorem 1
Given a graphon W, there exists a sequence \((W_n)_n\) such that
Theorem 2
The space is compact.
This implies that the space of graphons endowed with the cut metric is complete: Indeed, it is the completion of the space of finite graphs when itself is equipped with the cut norm.
While these results give us a nice framework to analyze graphs when their number of vertices grows very large, an issue arises when we consider the limit of a sequence of sparse graphs. The (edge) density of a graph can be defined as the fraction of its number of edges \(|E(G)|\) with respect to the maximum possible, which for simple directed graphs is \(|V(G)|(|V(G)|-1)\)
A dense graph is one such that for its two magnitudes \(|E(G)|\) and \(|V(G)|\) there is a relation like \(|E(G)| = O(|V(G)|^2)\), while for a sparse graph holds \(|E(G)| = o(|V(G)|^2)\). For sequences of graphs \((G_n)_n\), this translates into
The density of a graph can be rewritten as a sum
The form (38) of the density of a graph is a particular case of the p-norm for weighted graphs
where \(\alpha _i\) is the weight of the ith node, \(\beta _{ij}\) is the weight of the edge that connects nodei with nodej (with \(\beta _{ij} = 0\) if the two are not adjacent) and finally \(\alpha _G\) is the total weight of the graph
Now we see that when we consider an unweighted, directed, simple graph, Eq. (38) is Eq. (39) when p equal 1. We may define the p-norm for graphons, too: For \(1\le p < +\infty \), it has the form
while \(\Vert W\Vert _\infty \) is the essential supremum of W. Notice that \(\Vert W^G\Vert _p = \Vert G\Vert _p\) for all \(p \in [1,+\infty ]\). This tells us that a sequence of sparse graphs will converge toward one of vanishing density, so that its limit graphon will effectively be the uninteresting null graphon.
To overcome this difficulty, we will consider the normalized graphs, i.e., we will divide them by their 1-norm: This will allow us to compare graphons with different densities and to consider meaningful sequences of sparse graphs, with the aim of introducing and studying sparse graphons.
Results in Borgs et al. (2014) assure that normalized graphs converge, up to subsequences, to an \(L^p\) graphon in the cut metric; so do sequences of step graphons. Moreover, the \(L^p\) ball of graphons is compact with respect to the cut metric (up to identification of objects at zero distance).
At last, we recall the concept of connectivity and connected component for a graphon.
Definition 1
Given a graphon \(W :\Omega ^2 \subseteq [0,1]^2 \rightarrow \mathbb R^+\), we say it is disconnected if there exists a set \(S \subset \Omega \) such that S has positive Lebesgue measure and
We say that W is connected if it is not disconnected. A subset \(S \subset \Omega \) such that S is connected and \(S \cup \{x\}\), for all \(x \in \Omega {\setminus } S\), is not, is called connected component of the graphon W.
Clearly, we can always write a graphon as a disjoint union of its connected components.
Definition 2
Given a graphon \(W :\Omega ^2 \subseteq [0,1]^2 \rightarrow \mathbb R^+\) and point x and \(y \in \Omega \), we define
as the in-degree of x and the out-degree of y with respect to the graphon W(x, y), respectively. The two definitions agree in case of undirected, i.e., symmetric, graphons.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Düring, B., Franceschi, J., Wolfram, MT. et al. Breaking Consensus in Kinetic Opinion Formation Models on Graphons. J Nonlinear Sci 34, 79 (2024). https://doi.org/10.1007/s00332-024-10060-4
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s00332-024-10060-4